From 42710ea3ce7f4af826120b341c4bf277604fe080 Mon Sep 17 00:00:00 2001
From: Adrian Garcia Badaracco <1755071+adriangb@users.noreply.github.com>
Date: Wed, 2 Sep 2020 18:01:37 -0500
Subject: [PATCH] fix example

---
 rfcs/20200902-pickle-for-keras/README.md | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/rfcs/20200902-pickle-for-keras/README.md b/rfcs/20200902-pickle-for-keras/README.md
index b25ef54ee..847ce227e 100644
--- a/rfcs/20200902-pickle-for-keras/README.md
+++ b/rfcs/20200902-pickle-for-keras/README.md
@@ -100,11 +100,13 @@ Solution (2) would look something like this (assuming `Model.save` worked with `
 class Model:
 
     def __reduce_ex__(self, protocol):
-        b = io.BytesIO()
-        self.save(b)
+        b = io.BytesIO()  # python built in library
+        self.save(b)  # i.e. tf.keras.Model.save
         return load_model, (b.get_value()),)  # where load_model is tf.keras.models.load_model
 ```
 
+Note how this exactly mirrors the aforementioned blog post regarding PyTorch ([link](https://matthewrocklin.com/blog/work/2018/07/23/protocols-pickle)).
+
 By implementing this in all of Keras' base classes, things will automatically work
 with custom metrics and subclassed models.